Spectral Recognition Techniques and MLC of IRS P6 LISS III image for Coastal Landforms Extraction along South West Coast of Tamilnadu, India

The Remote sensing technology is measured or observed reflected energy to construct an image of the landscape beneath the platform passage in a discrete pixel format. The geometric and radiometric characteristics of remotely sensed image provide information about earth's surface. In the present study, the primary data product obtained from IRS P6 satellite LISS - III images (23.5 m) are used to extract the landforms the South West coast of Tamil nadu, India. The study area comprises different types of landforms in nature. The selected image processing techniques are employed such as, geometric correction, radiometric correction for removal of atmospheric errors and noise from image and to identify spectral and spatial variations in structure, texture, pattern of objects in the image. Here, the spectral recognition statistics namely edge detection; edge enhancement; histogram equalization, principal component analysis and maximum likelihood classifier algorithms are applied for demarcate the coastal landforms. In the maximum likelihood classification process, the spectral properties (Digital Number) of an object (kn) in the image has been identified using mean and covariance of pixels in training set, then the probability function (Px) determines the distribution of that group of pixels (class) in the image. The coastal landforms are segmented as separate class from the image based on their spectral and spatial characteristics such as shoreline, beach, sand dunes, erosional and accretion lands, water body, river deltas, and manmade infrastructure with attribute of shape, area, location and spatial distribution.